README.md

---
model-index:
- name: parthiv11/stt_hi_conformer_ctc_large_v2
  results:
  - task:
      type: automatic-speech-recognition
    dataset:
      name: Librispeech (clean)
      type: librispeech_asr
      config: other
      split: test
      args:
        language: en
    metrics:
    - type: wer
      value: 8.1
      name: WER
language:
- hi
metrics:
- wer
library_name: nemo
pipeline_tag: automatic-speech-recognition
---
## Model Overview

This collection contains large size versions of Conformer-CTC (around 120M parameters) trained on ULCA & Europal with around ~2900 hours. The model transcribes speech in Hindi characters along with spaces for Hinglish speech.

## Model Architecture

Conformer-CTC model is a non-autoregressive variant of Conformer model for Automatic Speech Recognition which uses CTC loss/decoding instead of Transducer. You may find more info on the detail of this model [here](https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/main/asr/models.html).

## Training

The NeMo toolkit [3] was used for training the models for over several hundred epochs. These model are trained with [this example script](https://github.com/NVIDIA/NeMo/blob/main/examples/asr/speech_to_text_bpe.py) and [this base config](https://github.com/NVIDIA/NeMo/blob/main/examples/asr/conf/conformer/conformer_ctc_bpe.yaml).

The tokenizers for these models were built using the text transcripts of the train set with [this script](https://github.com/NVIDIA/NeMo/blob/main/scripts/tokenizers/process_asr_text_tokenizer.py).

The checkpoint of the language model used as the neural rescorer. 

### Datasets

All the models in this collection are trained on Hindi labelled dataset (~2900 hrs):

- ULCA Hindi Corpus
- Europal Dataset

## Performance

The list of the available models in this collection is shown in the following table. Performances of the ASR models are reported in terms of Word Error Rate (WER%) with greedy decoding and 6-Gram KenLM trained on AI4Bharat Corpus and Europal.

| Decoding        | Version | Tokenizer           | Vocabulary Size | MUCS 2021 Blind Test* | IITM 2020 Eval Set | IITM 2020 Dev Set | Common Voice 6 Test* | Common Voice 7 Test* | Common Voice 8 Test* |
|-----------------|---------|---------------------|-----------------|------------------------|--------------------|-------------------|----------------------|----------------------|----------------------|
| Greedy          | 1.10.0  | SentencePiece Unigram | 128           | 9.37%/2.74%            | 12.93%/5.60%       | 12.63%/5.49%      | 13.16%/4.5%          | 13.5%/5.2%           | 14.37%/5.95%         |
| 6-Gram KenLM**  | 1.10.0  | SentencePiece Unigram | 128           | 11.79%/3.35%           | 15.96%/6.39%       | 15.49%/6.25%      | 17.05%/5.43%         | 17.77%/6.23%         | 19.18%/7.1%          |

*- Normalized and without special characters and punctuation.

**- KenLM with 128 beam size with n_gram_alpha=1.0, n_gram_beta=1.0.

## How to Use this Model
- Can also be used from NGC, intrution [here](https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/stt_hi_conformer_ctc_large).
- Follow [colab](https://colab.research.google.com/drive/1mLWVCbe4JFnooDoQLG0_33Je0LXdCZjO?usp=sharing) to use it directly


### Input
This model accepts 16000 KHz Mono-channel Audio (wav files) as input.

### Output
This model provides transcribed speech as a string for a given audio sample.

### Licence (Credit goes to Nvidia)
License to use this model is covered by the [NGC TERMS OF USE](https://ngc.nvidia.com/legal/terms) unless another License/Terms Of Use/EULA is clearly specified. By downloading the public and release version of the model, you accept the terms and conditions of the [NGC TERMS OF USE](https://ngc.nvidia.com/legal/terms).